OpenText.org: the problems and prospects of working with ancient discourse

1. Introduction The vast majority of studies in corpus linguistics have focused upon contemporary usage of modern languages. However, although there have been a number of studies of the earlier periods of some of these languages, such as Old English and Old French, they have tended to adopt the methods developed for modern languages. In our theoretical paper (Porter and O'Donnell 2001), we have explored the methodological challenges and questions posed by the study of an epigraphic language, such as Hellenistic Greek. 1 In particular, we have found the need for closer attention to the criteria used in the compilation of a corpus, the integration of the levels of annotation applied to a corpus, and maintaining a focus upon both traditional referential access and narrative or sequential text-analysis. 2 As a result, our approach is more than simply computer-aided text analysis, but in some ways fulfils the goals of the originators of corpus-based linguistics. It does this by performing a full analysis of the language, utilizing a structured corpus, rather than analyzing just a small portion of a much larger corpus. The textual orientation of classical and New Testament scholarship is compatible with and, in fact, requires a micro-pattern analysis, and the reading and analysis of full texts. Rather than placing it outside the scope of corpus linguistics, we argue that this perspective offers new avenues for the discipline, ones that we are exploring in OpenText.org. OpenText.org is a web-based initiative dedicated to creating resources for the linguistic analysis of Hellenistic Greek, and especially the Greek of the New Testament, in collaboration with interested scholars around the world. OpenText.org aims to make use of the insights and methods of corpus linguistics, specifically in terms of building a representative corpus of Hellenistic Greek, richly annotated according to a functional discourse model. This bottom-up discourse model relies upon the notion of levels of formal analysis. We parse all forms, beginning with morphology, but categorize from the word group up. We have found that the word group constitutes the smallest meaningful unit for discourse analysis. The increasingly higher levels are those of the clause, paragraph and discourse. One of the major principles of our annotation scheme is to mark features at the level of discourse at which they function. This has required the development of level-reflective and level-sensitive notational categories, as certain elements such as a conjunction may operate either at the word …